Impact of Educational Level on Study Attrition and Evaluation of Web-Based Computer-Tailored Interventions: Results From Seven Randomized Controlled Trials

BACKGROUND: Web-based computer-tailored interventions have shown to be effective in improving health behavior; however, high dropout attrition is a major issue in these interventions. OBJECTIVE: The aim of this study is to assess whether people with a lower educational level drop out from studies more frequently compared to people with a higher educational level and to what extent this depends on evaluation of these interventions. METHODS: Data from 7 randomized controlled trials of Web-based computer-tailored interventions were used to investigate dropout rates among participants with different educational levels. To be able to compare higher and lower educated participants, intervention evaluation was assessed by pooling data from these studies. Logistic regression analysis was used to assess whether intervention evaluation predicted dropout at follow-up measurements. RESULTS: In 3 studies, we found a higher study dropout attrition rate among participants with a lower educational level, whereas in 2 studies we found that middle educated participants had a higher dropout attrition rate compared to highly educated participants. In 4 studies, no such significant difference was found. Three of 7 studies showed that participants with a lower or middle educational level evaluated the interventions significantly better than highly educated participants (“Alcohol-Everything within the Limit”: F (2,376)=5.97, P=.003; “My Healthy Behavior”: F (2,359)=5.52, P=.004; “Master Your Breath”: F (2,317)=3.17, P=.04). One study found lower intervention evaluation by lower educated participants compared to participants with a middle educational level (“Weight in Balance”: F (2,37)=3.17, P=.05). Low evaluation of the interventions was not a significant predictor of dropout at a later follow-up measurement in any of the studies. CONCLUSIONS: Dropout attrition rates were higher among participants with a lower or middle educational level compared with highly educated participants. Although lower educated participants evaluated the interventions better in approximately half of the studies, evaluation did not predict dropout attrition. Further research is needed to find other explanations for high dropout rates among lower educated participants.